Thanks for the quick response! I know of course that different TV's are going to scan different areas, but there are still engineering standards. You need a reference point. You would color correct on a calibrated NTSC monitor, even though you know everybody's tv set will show something slightly (or awfully!) different, right? So what I'm after again is the specifications like "640x480" or something for TV transmitted, safe, title. Then, a fellow could shrink the image to fit the TV screen side-to-side to spec. If you shrink to transmitted, I don't believe anything other than a studio monitor with underscan will be seeing black on the sides. So you lose a little, but not the sick amount you do now. I'd like to see all the work you know? It's not a question of "fixing it in post" I shoot a movie in 1.85 or 2.40 or whatever for the theater. That's the composing aspect. It should be faithfully reproduced on the little screen, especially for my purposes, as I'm a cinematographer!
I wonder what they do for commercial DVD's. They're 720x480. Are we missing the sides on 4x3 TV's? I don't think they shrink the image inside because they come up in the computer with perfect borders. Blast! I just want a true 1.85 easily!

I'm far from an expert on this subject but I do know that it is a massive issue particularly if your shooting film and then showing it on a TV system. From memory film starts out 3:2 and the telecine will crop it to 4:3 so you've already got a safe area problem there. To further confuse the issue most film today is shot at around 2:1 so things get even worse, sometime its necessary to pan during transfer so action is not lost.

With commercial DVDs, well thats another issue, they may decide to do it 16:9 or 2:1 which means on a 4:3 TV its awefully letterboxed, on my 16:9 set there is still some letterboxing however SOME DVDs have two video tracks, with different aspect ratios. And some of the top range DVD players will do ARC.

But basically your right, your TV is normally masking a bit off all sides of the image. One reason for this is instability in the blanking period, without the masking you get jittery edges. It gets much worse on VHS, try recording onto a VHS tape and then bring it back as DV and you'll see what I mean.


Depending on the nature of the film they may also run it through an aspect ratio converter and transmit a 14:9 image as thats a reasonable compromise.

It all looks just fine on the PC monitor because your just seeing DV pixels mapped to screen pixels and there's no adjustments being made. Safe area is only one of the reasons for monitoring on an analogue monitor.

I'm certain you can find some more definative information from someone like SMPTE on just what the safe areas are supposed to be. Regardless though I don't think you'll ever avoid loosing some part of your image. Even if you zoomed your image into the safe area to ensure the audience will always see everything that you intended you are wasting resolution, you've now got pixels that could contain part of your image having nothing in them. I guess thats something you've got to weigh up.

Something that has been talked about at length here in the forum - might want to do a search, not because I'm too lazy to write it out, but it's one of those things that are sometimes hard to get your head around, and one of the posts/explanations might sort of click better then others for you. That said, here goes as short as I can do it...

You are not going to get full width in an absolute sense if everything works right, because the picture is too wide and too tall for the TV by design. If you do resize or something to get full width, then on some TVs you will see black bars on the sides of the picture.

In your case, when you recorded your film to DVCam, you should have gotten the full width (or as close to it as your equipment allows). If you treat it as DVCam throughout the process, and go back out to DVCam, everything should be as it was originally. In other words, you want to import the clip(s) to the timeline, rightclick on the clip(s) selecting properties, go to the 2nd tab, and make sure everything's as it's supposed to be for the imported footage. Then go to the proj. settings, and make sure everything agrees, and finally when you render, again make sure everything is the same: pixel aspect ratio, size, field order, that sort of thing. When you play back your DVCam footage from whatever hardware device, should be as the original you transferred.

Put another way, if you captured footage to DVCam that was letterboxed widescreen, and it wound up on your PC as 720 x 480 DV that appears to have squished width, keep everything that way throughout the editing process and on playback the player should expand the width as nec. When you have something like SVCD &/or DVD mpg2 with the widescreen flag set, the TV itself will show black bars top and bottom, often in addition to what was recorded in the 720 x 480 frame.

Where you get distortion and such is when you start changing things around - the only time you should have to is when converting formats &/or you're adding material that doesn't match your DVCam source. If you're converting formats, say going from DV to DVD, templates often do all the work for you. In any case, usually best off IMO to keep everything DVCam as before, just select your destination properties on render by going through the custom dialog - during render you should see your picture in the preview window at the proper aspect. The only watch out, is if it doesn't look as you want during render, you might have to go back to the clip properties or the switches dialog and turn off maintain aspect ratio - if for example you had anamorphic DVD footage & wanted to spread it out to full frame width on render. Going to mpg2 you might set the aspect ratio, which is a flag telling the player to expand things on playback. Most players should know DV & DVCam is slightly narrowed and blow it up. It's pretty common sense so no worry -> if you get black bars where you shouldn't, uncheck maintain aspect ratio generally and check the aspect settings for render, and again, only should be needed when changing formats.

Adding non-confirming footage gets more complicated, but only in concept. The easiest way I know to explain it is to import a few still into a DVCam proj to get a feel for what's happening. Assuming your proj. fits the std DV template at 720 x 480, import a still of that size to the timeline. Notice how it displays, and render the image (render just selected range as nec) to DV and see how it plays back. Then right click on your image(s) on the timeline, select properties, and set the aspect ratio to match your DV footage. See how it displays and renders, as you just told Vegas that the image already conforms to spec. and doesn't have to be altered.

A notable exception is with DVDA, where you might see black bars in the preview but not on viewing the rendered proj..

RE: the frame sizes themselves, you might see 960 x 540, which is DVD widescreen anamorphic video expanded as intended by a player. Some broadcast type equipment sticks with the perhaps more correct 540 height giving you overscan for NTSC TV in that dimension. Trivia, some capture equipment years ago used to go after the 540 height. As it's far less critical then the width dimension, full size is most often seen at 480 height nowdays. [The extra width is critical because of variance among picture tubes, because it hold additional info with broadcast TV, because of the signal characteristics, all sorts of trivia you probably won't have to bother with]

The standard NTSC full width is 720, the cropped full width is 704 (some of the info mentioned above stripped out), and the basic needed to fill a TV screen is 640, which conforms to the 4:3 ratio on your average PC monitor. Poke around the site dvdrhelp.com and you'll find an article on CVCD which contains some interesting info on the actual effective width, perhaps giving insight into the SVCD spec of 480 width. You have 352 x 480 (half resolution giving both fields), and I've seen less commonly 355. Quarter size cuts the height in half to one field, but will still fill a TV screen. As you can see the average TV is quite forgiving, and terms are often used as catchalls rather then correctly.

Hope something there helps...

So what you're asking is why 720x480 isn't letterboxed when you play it on a tv, correct? 720x480 is the DV NTSC specs. It's like the SD NTSC spec of 640x480. That will appear in the "4x3" ratio (that refers to the ratio of your TV, not your footage). If you want to make it appear wide screen (by manualy putting black bars in it), just plop a png (o jpeg, bmp, psd, etc) ontop of it with the bars in the correct place (don't forget to chroma key or alpha channel) and you'll have wide screen. You will notice some of the picture is cut out though.

If you want to take your 720x480 and have it be wide screen like you view on a DVD, just render it with a different asped ratio. That should do it. I've tested this before (but with SVCD) from movie trailers i've downloaded from the net, and if i render it with the wider asped ratio it will be wide screen on TV, and if i use the normal aspect, it will squish my widescreen stuff.

OK. So here's what I came up with, for those who seek the same. First, let me succinctly restate the goal: To see all of the 1.85, or 2.40, or 16x9, or DV(1.5), when presented on a TV-set. Why? Because When I film a movie in 1.85, or Stirraro does, or Mickey the Clown does, the shot is composed for that frame in the viewfinder. A lot of thought and choices go into that frame, and that's what I want to see reproduced. When you go to a movie they don't cut off the damn sides. Obviously It'll have to be letterboxed on a 4x3 TV! That doesn't bother me. More importantly, I get jobs based on my reel, my ability to light and choose composition for the frame. The director wants to see what my choices were. It is usually viewed in a production office on a TV. I don't want to show "kinda what we shot"! Man we spend millions making that frame. Directors and DP's and Production Designers and all the set strives to make that as good as it can be...
-for the rate;) -
Hell, if the producer knew we were only going to see the middle part of the picture he'd ask us to reduce our rate per-the viewable area!


The result: TV is 4/3=1.33 aspect ratio. 640/480=1.33. So 640= the width. But that's if the pixels were square. Since the dv pixels are rectangular - .9091 as wide as they are tall- you'd need 704 of them to be as wide as 640 square pixels. So, 704x480 is all that's possible to see on a (regular)TV. So use the "pan/crop"1.85 preset, then "track motion" to size down to 704. Or you can do it all in pan crop by entering 726(=720+16)x346 (=1.85) in the "size" boxes .I'd have to think that if you wanted your whole frame to be seen on a DVD on a regular TV, you'd at least reduce to this. Obviously I'm not talking rendering to these dimensions, but reducing the image area to a smaller dimension so there is no waste.

What , if anyone knows, are the realistic display dimensions of a TV image? The area that effectively makes it onto the screen? It seems more like 670x447. Anyone?

"If you want to make it appear wide screen (by manualy putting black bars in it), just plop a png (o jpeg, bmp, psd, etc) ontop of it with the bars in the correct place (don't forget to chroma key or alpha channel) and you'll have wide screen. You will notice some of the picture is cut out though."

IMO an easier way is to use pan/crop, reducing the size of your frame, then keep the proj at full height -> the bars are generated on render.

Trivia (couldn't resist): 35mm film is 4:3 cut off to ~16:9 which started as a way to make the movie in the theater different from the much feared TV way back when...

That irrelevant comment made, please try and forget about rectangular pixels and such, or at least try to think of them as an imaginary measurement -> focus on results. IMO, start with the output format, and work backwards -> if you need 720 x 480 for DVD as an example, the choices are already made. If you're going broadcast, then in most cases you're pretty limited to what is legal, and/or what the client wants [some are archiving in near HDTV but doing a pan/scan before broadcast today].

FWIW, If you're showing whatever on an the average TV, you will need at least 704 width, though 720 is generally better for compatibility with everything else. And you will lose most of the difference between 640 and 704 or 720. Common practice with commercial DVDs is to first decide on a frame size (16:9 is a catchall term in practice). Then the video is encoded at 720 x 480... You might have an actual picture area of 853 width x 360 height, which is common, and black bars are added top and bottom. [you can do this in Vegas *IF* you're careful -> might have to play around a bit with the settings, and might have to do a pre-render, but generally if you check the box on the save as dialog to stretch the video it will do the right thing -> this is counter-intuitive and the preview will not match the output, so you *NEED* to run a test or two.]

The result ideally will be a video that appears distorted, shrunken horiz. in an editor. There are flags that are set in the mpg2 file: one of these you can set in Vegas for the 16:9 ratio, another for horiz size (often 540) has to be set in something like restream (requires a de-muxed file so render to streams if you use it). The resultant mpg2 when played should be displayed at 960 x 540. Take most any widescreen DVD release apart, and this is pretty close to what you'll find. Play the DVD on the average TV, and there are parts that will not be seen compared to what you get on your PC monitor.

The only standards that I'm aware of that allow whatever frame size you want are thing's like mpg4, DiVX, wmv etc... Mpg4 and/or DiVX are playable today by some DVD players, and if one believes MS, wmv will follow shortly.

Where you would worry about pixel ratios and if they're square or not, is usually along the lines of mixing content, or if you're trying anamorphic wmv (IMO interesting, but far from proven).

Kirk,

Televisions all overscan to different degrees. Professionals frame their shots with this in mind. That's why we have "safe action" and "safe title" areas. It's the way it works, get used to it.

Just frame your "multi-million dollar" shot a little looser so that the area of interest in the shot is at least within the "safe action" area. All of the professionals I know (including myself) are fully aware of the overscan issue and compensate accordingly. I suggest you start doing that and quit banging your head against that wall.

John

"When you go to a movie they don't cut off the sides."

Kirk, i'll have to reject this premise. It seems like most every time i see a movie in the theater lately, there's at least a foot or two of the movie just barely visible on each side of the black frame around the screen, and maybe a few inches above and below too. They are cutting off some of the sides. I'm guessing it's because the screens are set up for 1.85 and when they project anything wider than that, they feel the movie-going audience would object to having the picture not fill the frame ... so they crop.

First, thanks for all the info and time taken. Wow John, don't take it personally! What you say is true of course, and when I shoot commercials or TV shows, I frame for TV safe. I shoot a framing chart, and in telecine, I fine tune the sizing. Standard practice. So on a perfect monitor(!) you see what you composed for. If you lose image, at least it's symmetrical. If you shoot in a theatrical aspect, perhaps "professionals like you" could show me the new 1.85 safe/transmitted lines are on the ground glass! (couldn't resist, I'm chuckling, not being mean) You shoot it, you project what was seen in the viewfinder. I stress once again that a lot of people are working to put meaning into that frame, that care about their job and the content. Why wouldn't you want to see it if you could? Just shrink the image a little so the sides aren't outside the picture area.

Regardless, TV underscans. Thanks gain for the help. It appears that 704x346 is the biggest image size of a 1.85 aspect that will be displayed edge to edge in a TV monitor. The 720x480 DV or DVD render size doesn't change.

"It appears that 704x346 is the image size of a 1.85 image that will be displayed edge to edge in a TV monitor."

Perhaps, but not necessarily, the amount of overscan varies from TV to TV. The fact of the matter is that you have no control over what type of TV your project will be viewed on. I suppose that you could adjust for absolute worst case, but then people with reasonably well adjusted TV's would see black bars on the left and right of the image. If the viewer has a TV that overscans severely, it's their problem, not yours.

I spent years as a recording engineer and went to great lengths to make sure everything sounded great but, once the recording was released, it was out of my hands. That pristine recording could potentially be played on a boom-box with the bass cranked all the way up. Obviously, that wasn't the way I intended for it to be played, but there was nothing that I could do about it. Except for rendering the image so small that it would be guaranteed to show the entire frame on even the most misadjusted TV's, there isn't much you can do about it either. While I will take it into consideration, I refuse to cater to the absolute lowest common denominator.

John

I believe there is an (up/down) adjustable hard matte in the gate of the projector. It is however committed physically to 1.85. How the theater's screen is set up relative to the projector....

"the amount of overscan varies from TV to TV" As have many of the postings including mine recognized. The math dictates that if an image is 1.33 or 4x3, and it's 480 high, it should be 704 wide, taking rectangular pixels into account. That's best case scenario though right? That's why I'd like to know what TV SAFE size is. Armed with that, I think an educated decision could be made, in much the same way as you said, framing, or in this case sizing, the image with this in mind.

You guys are working too hard at this. If you want to see everything as you shot it, just reduce the image slightly using the Track Motion tool. About 10% would do it. You can adjust for slightly more or less after seeing the results. I do it all the time when I realize in editing that I really need to show the entire frame of a particular shot.

By the way: My Sony camcorder uses a slight overscan on both the viewfinder and swing-out lcd, I assume so that the user doesn't capture something that a TV monitor won't show.

Also, Kirk, DV is not 4:3. It's actually slightly wider than that. That explains why, if you import a 4:3 still photo onto the timeline, you'll see narrow black bars on the left and right sides.